ORBiS Tree

OPEN AI KNOWLEDGE WIKI · ENGLISH

AI, explained as
a connected map.

Understand models, concepts, organizations, products, research, and infrastructure without getting trapped in isolated definitions. Start with a clear explanation, verify the sources, then follow the next concept.

40 English articles Free to read Open Korean community edition →
English edition
These pages are curated translations of the Korean source articles. Each translation has its own indexable URL and is linked to the original with hreflang metadata.
How the edition works →

START HERE

Five useful entry points

KNOWLEDGE MAP

Browse by topic

All topics →

RECENTLY UPDATED

English articles

EN

Datasheets for Datasets

A documentation framework for describing how datasets were created, composed, maintained, and intended to be used.

#Dataset#Datasheets#Documentation#Transparency
EN

System Card

Documentation describing a deployed AI system, including capabilities, evaluations, mitigations, and limitations.

#Documentation#Governance#System Card#Transparency
EN

Safety Evaluation

Testing designed to measure risks, harmful behaviors, and the effectiveness of safeguards in an AI system.

#AI Safety#Evaluation#Risk#Safety Evaluation
EN

Red Teaming

Adversarial testing intended to discover failure modes, unsafe behavior, or exploitable weaknesses in an AI system.

#Evaluation#LLM#Red Teaming#Safety
EN

Data Provenance

Information about where data came from, how it was collected, and how it changed before use.

#Data Provenance#Dataset#Governance#Transparency
EN

Benchmark Data Contamination

The presence of evaluation examples or closely related material in training data, potentially inflating benchmark results.

#Benchmark#Data Contamination#Evaluation#LLM
EN

Model Card

A structured document that describes a model's intended use, evaluation, limitations, and other important context.

#Documentation#Evaluation#Model Card#Responsible AI
EN

Latency and Throughput

Two core serving metrics describing response delay and the amount of work a system completes over time.

#Benchmark#Inference#Latency#Throughput
EN

GPU

A highly parallel processor widely used to train and run modern machine-learning models.

#CUDA#GPU#Hardware#Inference
EN

Context Window

The amount of tokenized input and generated text a model can consider within one inference context.

#Attention#Context Window#LLM#Tokenization
EN

KV Cache

A cache of attention keys and values used to avoid recomputing earlier context during autoregressive generation.

#Inference#KV Cache#LLM#Transformer
EN

Inference

The process of running a trained model on new input to produce predictions, embeddings, or generated output.

#Deployment#Inference#LLM#Serving